EIB Papers Volume 13. n°1/2008 - European Investment Bank

High variability of the
public-capital stock
reduces its long-run
impact on output.
Figure (4b) shows that a high ratio of public capital to private capital is negatively and significantly
related to the impact of public capital on output (the t-statistic is 2.22, with p = 0.038). Apparently,
the impact of public capital does not depend on its absolute level, but on its level relative to private
capital. If the public capital stock is large relative to the private capita stock, the long-run impact of
public capital on output is lower.
The regression line in Figure (4c) does not yield a significant relationship (the estimated t-statistic is
0.19, with p = 0.852). So the diversity in our sample with respect to the change in the public-capitalto-GDP
ratio is not related to our results for the long-term impact of public capital on output. In
other words, this finding suggests that there is not a systematic difference with respect to the longrun
impact of public capital on output between countries that saw their capital-to-GDP ratio decline
and those that saw this ratio increase.
Finally, Figure (4d) suggests that there is a negative and significant relationship between the
variability of the public capital stock and our findings for the long-term impact of public capital on
output. The estimated t-statistic is -2.40 (p = 0.027). So these findings suggest that high variability of
the public capital stock reduces the long-run impact of public capital on output. 11
Figure 4. Government capital and its long-run effect on output, 1960-2001
Estimate of LR e�ect on GDP
72 Volume13 N°1 2008 EIB PAPERS
3
2
1
0
-1
-2
-3
POR
ESP
CAN
(a) Level of public capital
NOR
GRE
FRA USA
SWE FIN
AUS
DEN
BEL
ITA
SWI
ICE
GBR
0.2 0.4 0.6 0.8 1.0
Government capital / GDP
NET NZL
Note: The vertical axis shows the long-run impact on GDP of a shock to public capital, while the horizontal axis shows the
average public-capital-to-GDP ratio for the country concerned.
11 In principle, there may be two sources of high standard deviations in the public-capital-to-GDP ratio, (i) public investment
being “erratic”, and (ii) public investment having a strong trend such that it is very different at the end of the sample from
its initial level. As the correlation between the absolute change and the standard deviation of the public-capital-to-GDP
ratio is 0.79, we conclude that the variability in the public capital GDP to ratio is largely driven by the second source.
AUT
IRE
JAP